CGENet: A Deep Graph Model for COVID-19 Detection Based on Chest CT

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Accurate COVID-19 diagnosis is crucial for disease control.
• Chest CT scans are vital for identifying COVID-19.
• Existing diagnostic methods require improvement in speed and accuracy.

Purpose of the Study:

• To develop a novel, explainable AI system (CGENet) for COVID-19 diagnosis using chest CT images.
• To enhance diagnostic accuracy and efficiency for COVID-19 detection.
• To provide visual explanations for AI-driven diagnoses.

Main Methods:

• Proposed CGENet system integrating graph embedding and extreme learning machine (ELM).
• Utilized an optimal backbone selection algorithm based on transfer learning.
• Incorporated graph theory with ResNet-18 and k-nearest neighbors.
• Trained ELM as the CGENet classifier.

Main Results:

• Achieved an average accuracy of 97.78% on a large COVID-19 dataset via 5-fold cross-validation.
• Demonstrated the system's effectiveness in distinguishing COVID-19 from other conditions.
• Generated Grad-CAM maps for visual interpretability of diagnoses.

Conclusions:

• CGENet is an effective and efficient tool for assisting COVID-19 diagnosis.
• The explainable nature of CGENet enhances trust and clinical utility.
• This AI-driven approach shows promise for rapid and accurate disease detection.